| dc.description.abstract |
Dengue fever, which is primarily transmitted by Aedes aegypti and Aedes albopictus, is responsible for approximately 390 million infections globally per annum. The current study was conducted to characterize the dengue risk within the districts of Colombo and Kandy. Information on socio-economic and demographic status, and knowledge, attitudes and practices (KAP) were gathered from 1000 randomly selected patients and 1000 non-dengue reported households within each district using an interviewer administered questionnaire. Routine entomological surveillance was conducted from February 2016 to July 2018 at monthly intervals within 10 high risk Medical Officer of Health (MOH) areas in both districts. Monthly vector indices, [Premise Index (PI), Container Index (CI) and Breteau Index (BI)], meteorological parameters (monthly total rainfall, minimum and maximum temperature and relative humidity), landuse practices and socio-demographic data from all the 39 MOH areas relevant to the period of January, 2012 to December, 2018 were collected as secondary data. The socio-economic attribute. were statistically analysed by using the Chi-square test of independence and cluster analysis. Receiver Operating Characteristic (ROC) curves analysis was used to develop thresholds for dengue epidemic management. Principal Component based Linear Regression (PCLR) and Principal Component based Poisson Regression (PCPR) approaches were used to develop a spatial risk characterization model for dengue, while Seasonal Autoregressive Integrated Moving Average (SARIMA) approach was used to develop a temporal dengue prediction model. The climate change vulnerability of the local communities to dengue was evaluated by using the composite index method. Significant differences were identified among the test and control groups for basic demographic factors, living standards, knowledge, attitude and practices. The test group indicated similar risk factors, while the control group also shared more or less similar characteristics as depicted by the findings of cluster analysis. Further, improvement in key infrastructural facilities such as urbanization and waste collection, community education, public motivation, coordination and integration of control programmes, were recognized to be vital. Only PI and BI for Ae. aeopti (Blagp) were significantly associated with dengue epidemics at lag periods of one and two months. Based on Ae. aegypti, average threshold values were defined for Colombo as Low Risk (Blagp ≤2.4), Moderate Risk (3.8 ≤ Blagp ≥5), High Risk (Blagp ≥ 5), along with Blagp ≤ 3.0 (Low Risk), 4.2 ≤ Blagp < 5.3 (Moderate Risk) and Blagp ≥ 5.3 (High Risk) for Kandy. Further, PI ≤ 5.5, 8.9 ≤ PI ≥ 11.9 and PI ≥11.9 were defined as Low Risk, Moderate Risk and High Risk average thresholds for PI in Colombo, while PI ≤ 6.9 (Low Risk), 9.1≤ PI ≥ 11.8 (Moderate Risk) and PI ≥ 11.8 (High Risk) were defined for Kandy. The best fitting model converged by PCPR was the best risk characterization model, with higher levels of goodness of fit indicators such as R2 and Adjusted R2 values of 90.08% and 89.88%, with an AIC value of 205.86. The best fitting forecasting models fitted for Colombo and Kandy are SARIMA (0,1,0) (3,0,0) and SARIMA (2,1,2) (1,0,0)12 respectively. Colombo Municipal Council MOH area had the highest vulnerability (0.49: moderate vulnerability) to dengue, while the Galaha MOH showed the lowest (0.15; very low vulnerability). KEYWORDS: Dengue, Risk prediction, GIS, Spatial model, threshold. |
en_US |